Guide ยท rally & garrison
How to build a rally that wins the trade
Most players build a rally the same way: stack the meta leaders, pile on attack joiners, send more troops. Almost all of that is wrong. A rally is a two-way trade decided by a hidden multiplier called SkillMod, and the biggest lever is which four joiner skills land in it, not your stats and not your numbers. Here is how it actually works, and how to build your comp for the server you are on.
- A rally is a trade, not a race. You win by destroying more of them than they destroy of you. Army size feeds a square root, so sending more troops has sharp diminishing returns.
- Every fight runs two SkillMods. Your hit is your offense over their defense; their counterattack is their offense over your defense. The skills on the bottom directly divide the ones on top.
- Stacking attack is the trap. Four attack joiners hit hard once, then eat the full return blow. Read which half of the enemy is loaded and bring the opposite half of yours.
- Stats never pick your joiners. SkillMod is built from skill effects only, zero stats feed it. Whale upgrades change the final numbers, never which joiners are correct.
- Build by server age. Young server: stack Saul and Hilde, they are never wrong. Experienced server: your R4 assigns joiners and controls who gets counted.
The reasoning for each is below, for whoever wants it.
What actually decides a rally?
One line of math decides every fight:
damage = army_factor × att_per_troop / def_per_troop × type_bonus × SkillMod × fatigue / 100 deaths = ceil(damage)
There are two layers feeding it, and they behave completely differently.
The stats layer is everything in your battle-report stat panel: troop base, tech, gear, leader percentages, pets, buffs. It builds the per-troop values, and each side is a product:
att_per_troop = Attack × Lethality def_per_troop = Defense × Health
Because it is a product, balance beats a lopsided total: 100 × 100 = 10,000 beats 150 × 50 = 7,500. So inside your attack you want Attack and Lethality balanced, inside your defense you want Defense and Health balanced. Confirmed. But that is a gear, tech and charm question, not a skill question.
The SkillMod layer is separate, and it is where every hero and joiner skill lives. It never touches the stat panel. Breaking a garrison is not about wiping it to the last soldier; it is about the trade, destroying a larger fraction of them than you lose of your own, and SkillMod is the lever you actually control at fight time. The rest of this page is that layer.
Do you need more troops than the defender?
No, and this is the most expensive misconception in the game. Army size enters through a square root, and the smaller of the two total armies is locked in at the start of the fight:
army_factor = sqrt(your_squad_size × armyMin) armyMin = min(your total army, their total army), fixed at fight start
A square root means doubling your army does not double your hitting power. With labelled example inputs:
-- same defender both times, armyMin = 78,872 squad 40,000 → army_factor = sqrt(40,000 × 78,872) = 56,168 squad 80,000 → army_factor = sqrt(80,000 × 78,872) = 79,434 ratio = 79,434 / 56,168 = 1.41 (not 2.0)
Doubling the squad bought you +41% damage, not +100%. Confirmed. Past parity, extra bodies give sharply diminishing returns and the defender still grinds them down on the counterattack. The levers with real leverage are per-troop power and skills.
Which troop ratio should you send?
Start from what you are actually hitting, because the defender's mix is not random. Rally garrisons cluster around 60/40/0 or 60/20/20 (infantry / cavalry / archer): an infantry wall with cavalry behind it and few or no archers. A city you march on is different, closer to 35/25/40, with a real archer block behind the front.
The type triangle tells you what to answer them with: infantry beats cavalry, cavalry beats archer, archer beats infantry, for a flat +10% damage where you land it. Confirmed. So against a 60/40/0 wall, archers answer the infantry and your own infantry answers their cavalry; against a 35/25/40 city, cavalry answers the archer block. Two things stop this from being a clean formula, though. Targeting resolves front to back (infantry, then cavalry, then archers), so your squads do not always reach the type they counter before something else dies, and cavalry bypass and archer volley bend it further. That is why the exact split is roster-and-defender specific, and why the tool searches it instead of a fixed ratio.
The two SkillMods, and why stats do not feed them
SkillMod is the hidden multiplier where every hero skill, joiner skill and widget effect lives. It is a fraction:
SkillMod = (your DamageUp × your EnemyDefenseDown) / (their DamageDown × their DefenseUp)
Every skill drops into one of four buckets. Two raise your attack (damage up, shred their defense); two are defensive (cut their damage, raise your defense). And here is the part that changes everything: every fight runs two of these, not one.
your hit = (your offense) / (their defense) their counter = (their offense) / (your defense)
The bottom of each fraction is the direct counter to the top. Their DefenseUp divides your DamageUp; your attack-down divides their DamageUp. So one joiner's skill means nothing on its own: +50% damage up does nothing until you know what is in the denominator. You always pool both full sides, all three leaders plus up to four joiners each:
- Top of your hit: your leaders and joiners that push damage up or shred enemy defense.
- Bottom of your hit: their leaders and joiners that raise defense or cut your damage.
A defensive joiner helps the second fraction (their counter on you), an offensive joiner the first. The matchup decides which is worth more, never the joiner in isolation.
Stacking attack is the trap
You are attacking, so you stack attack heroes. It feels obvious and it is usually wrong, because of the two fractions above: pile everything into the top of your hit and the bottom of their hit stays empty, so you eat the counterattack at full force. Watch what swapping two of four attack joiners for one defensive and one enemy-debuff joiner does, using the calculator's hero values:
-- A: four attack joiners 2 Amane + 2 Chenko your hit ×2.25 your defense ×1.00 -- B: spread across families 1 Amane + 1 Chenko (damage up) 1.25 × 1.25 = 1.56 1 Saul (+Def, +HP) 1.10 × 1.15 = 1.27 1 Fahd (enemy attack down) 1.20 your hit ×1.56 your defense ×1.52
Option A hits about 44% harder on the opening blow, then takes the counterattack at full force. Option B hits softer but cuts that counter to about 66% of full (divided by 1.52). The two attack slots A gave up were only deepening a bucket it already had; they did nothing for the half of the fight where you take damage.
The sharper version of the same idea is reading the enemy and loading the opposite half:
They are offense-stacked
Everyone piling damage up. Their defense side is basically empty, so when you attack you already punch through. The real danger is their return fire. Answer with attack-down and damage-taken-down: that is the denominator of their attack on you, and it divides their fat numerator down. You win the trade.
They are a turtle
Defense up plus damage reduction. Your raw damage up gets divided away. Bring defense-shred: it lands in your numerator in a different bucket than their defense up, so they cannot soak it, and it diversifies your damage skills at the same time.
Build your comp by server age
You rarely know the enemy's exact skills, so in practice the meta does the scouting for you. People copy the meta leaders and almost always run the same joiners (Amane, Chenko, Howard, Gordon, Saul), so assume the popular list and counter it. How far you push past that scales with how sharp your server is.
- Young server, joiners still a struggle. Just stack Saul and Hilde in both attack and defense. They are jack-of-all-trades, never wrong, and you are already ahead of a kingdom stacking Chenkos or Howards.
- Some experience. Start differentiating your defense: swap your defensive joiners around. Join order does not matter, you pick exactly who is counted by shuffling who leads until the joiners you want are the ones that land.
- Experienced server. Plan it before the fight. Players are each assigned a joiner, everyone piles into the offensive rally, and you kick whoever is not needed at that moment; they rejoin after, five minutes is plenty. Who joins attack and defense is decided by an R4, not left to chance.
That last tier is where the R4 role actually earns its keep. You are not managing players, you are managing which skills land in the fraction. It is also the most fun part of the job.
How the simulator finds it
Every section above is a lever; the hard part is combining them against a specific defender with the specific heroes you own. Two tools do that.
- The live SkillMod calculator builds both sides of the fraction and shows both trades at once: your hit on them and their counter on you. Because stats never feed SkillMod, you do not need any real stat report to compare joiner sets here.
- Best Counter in the simulator takes a full defender and ranks every leader trio, joiner set and troop split your roster can field, scored by net casualty trade and a Monte-Carlo win rate. Use it when you want the actual win-or-lose result of a specific hold, where stats do matter.
What this covers
The simulator models KingShot Expedition rally versus garrison. You enter the active leaders and joiners directly, so what you simulate is the formation you specify; it does not model automatic garrison leader replacement when a higher-power player joins. It does not model Conquest or the Bear Hunt damage curve, which are different engines. Where a value is uncertain, the tool grades it rather than pretending, the same as this page.
FAQ
Do account stats affect which joiners I should bring?
No. SkillMod is built only from skill effects; zero stats feed it. Gear, tech, leader percentages and whatever the whales upgrade live in a separate layer that only multiplies in at the end. Because SkillMod is a clean multiplier, stat upgrades change the absolute result but never which joiners are correct, so you can test joiner setups without a real stat report.
Should I stack attack heroes for a rally?
Usually not. A rally hits both ways, and only two of the four skill families raise your outgoing damage; the other two shrink the counterattack. Four attack joiners hit harder on the opening blow but eat the full return, so reading the enemy and loading the opposite half of your roster trades better.
Which joiners should I bring to a rally?
Counter the enemy's loaded half: against an offense-stacked enemy bring attack-down and damage-taken-down, against a turtle bring defense-shred. On a young server where everyone copies the same list, Saul and Hilde are the safe all-rounders that are never wrong.
Do I need more troops than the defender?
No. Army size enters the damage through a square root and the smaller army is fixed at the start, so out-numbering has sharp diminishing returns. Per-troop power and skills decide more than raw numbers.
Related: the four SkillMod families · the damage formula · who counts in a rally · what to build next
Build your comp against a real defender
The SkillMod calculator compares joiner sets with no stats needed; Best Counter ranks the full rally your roster trades best with.